UTR Mrna Isoform by CRISPR–Cas9 Gene Editing Impairs Dorsal Root Ganglion Development and Hippocampal Neuron Activation in Mice

Downloaded from rnajournal.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press

Elimination of Calm1 long 3′′′′′-UTR mRNA isoform by CRISPR–Cas9 gene editing impairs dorsal root ganglion development and hippocampal neuron activation in mice

BONGMIN BAE,1,3 HANNAH N. GRUNER,1,3 MAEBH LYNCH,1 TING FENG,1 KEVIN SO,1 DANIEL OLIVER,2 GRANT S. MASTICK,1 WEI YAN,1,2 SIMON PIERAUT,1 and PEDRO MIURA1 1Department of Biology, University of Nevada, Reno, Nevada 89557, USA 2Department of Physiology and Cell Biology, University of Nevada, Reno School of Medicine, Reno, Nevada 89557, USA

ABSTRACT The majority of mouse and human genes are subject to alternative cleavage and polyadenylation (APA), which most often leads to the expression of two or more alternative length 3′′′′′ untranslated region (3′′′′′-UTR) mRNA isoforms. In neural tissues, there is enhanced expression of APA isoforms with longer 3′′′′′-UTRs on a global scale, but the physiological relevance of these alternative 3′′′′′-UTR isoforms is poorly understood. Calmodulin 1 (Calm1) is a key integrator of calcium signaling that generates short (Calm1-S) and long (Calm1-L)3′′′′′-UTR mRNA isoforms via APA. We found Calm1-L expression to be largely restricted to neural tissues in mice including the dorsal root ganglion (DRG) and hippocampus, whereas Calm1-S was more broadly expressed. smFISH revealed that both Calm1-S and Calm1-L were subcellularly localized to neural processes of primary hippocampal neurons. In contrast, cultured DRG showed restriction of Calm1-L to soma. To investigate the in vivo functions of Calm1-L, we implemented a CRISPR–Cas9 gene editing strategy to delete a small region encompassing the Calm1 distal poly(A) site. This eliminated Calm1-L expression while maintaining expression of Calm1-S. Mice ΔL/ΔL lacking Calm1-L (Calm1 ) exhibited disorganized DRG migration in embryos, and reduced experience-induced neuronal activation in the adult hippocampus. These data indicate that Calm1-L plays functional roles in the central and peripheral nervous systems. Keywords: 3′′′′′-UTR; calmodulin; alternative polyadenylation; mRNA localization; axon; dorsal root ganglion; hippocampus

INTRODUCTION neurons (Tushev et al. 2018). Thousands of genes in mouse and human express long 3′-UTR mRNA isoforms in brain Alternative cleavage and polyadenylation (APA) is the pro- tissues (Miura et al. 2013). For many genes, short 3′-UTR cess by which a pre-mRNA is cleaved and polyadenylated isoforms were found to be expressed across various tis- at two or more polyadenylation [poly(A)] sites. This com- sues, whereas the alternative long 3′-UTR isoforms were monly results in two (or more) mRNAs with the same pro- abundant only in brain. Our understanding of the in vivo tein coding sequence but different length 3′-UTR. APA functions of neural-enriched long 3′-UTR isoforms is limit- is pervasive, occurring in ∼51%–79% of mammalian genes ed despite the large number of genes shown to be affect- (Hoque et al. 2013; Lianoglou et al. 2013). The 3′-UTR is a ed in Drosophila, Zebrafish, mice, and humans (Smibert major target for posttranscriptional regulation via micro- et al. 2012; Ulitsky et al. 2012; Miura et al. 2013). RNAs (miRNAs) and RNA binding proteins (RBPs); thus, Multiple studies have examined the functions of 3′-UTRs harboring a longer 3′-UTR can theoretically confer addi- in vivo by generating mice that either genetically delete tional regulatory opportunities for transcripts (Sandberg parts of 3′-UTRs or introduce foreign poly(A) sites to cause et al. 2008; Mayr and Bartel 2009). Long 3′-UTRs impact production of mRNAs with truncated 3′-UTRs (Miller et al. translation in a cell context-specific manner (Floor and 2002; Perry et al. 2012). For example, a genetic approach Doudna 2016; Blair et al. 2017), and elements located in was implemented in mice to abolish the BDNF long 3′-UTR alternative 3′-UTRs can influence mRNA localization in isoform while maintaining expression of its short mRNA counterpart via the insertion of tandem SV40 poly(A) sites 3 These authors contributed equally to this work. Corresponding author: [email protected] Article is online at http://www.rnajournal.org/cgi/doi/10.1261/rna. © 2020 Bae et al. This article, published in RNA, is available under a 076430.120. Freely available online through the RNA Open Access Creative Commons License (Attribution 4.0 International), as described option. at http://creativecommons.org/licenses/by/4.0/.

1414 RNA (2020) 26:1414–1430; Published by Cold Spring Harbor Laboratory Press for the RNA Society Downloaded from rnajournal.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press

Loss of Calm1-L impairs neuronal function downstream from the proximal polyadenylation signal. has a specific function in the embryonic or adult nervous These mice displayed synaptic defects and hyperphagic system has not been addressed. obesity, ostensibly due to compromised mRNA localiza- Here, we utilized CRISPR–Cas9 to generate the first tion to dendrites and impaired translational control (An mouse lines lacking expression of a long 3′-UTR mRNA iso- et al. 2008; Liao et al. 2012). The recent advent of form while maintaining normal expression of its short 3′- CRISPR–Cas9 gene editing has revolutionized the speed UTR APA counterpart. We uncovered that Calm1-L levels and efficiency of generating deletion mouse strains were high in the DRG and hippocampus, and its expres- (Wang et al. 2013). This presents an exciting new opportu- sion was largely restricted to neurons. In vivo phenotypic nity for rapidly generating 3′-UTR isoform-specific knock- analysis of mutant embryos revealed disorganized DRG out mice. CRISPR–Cas9 has been implemented to axon and cell body positioning, whereas adults displayed successfully generate mice with a deletion that removed a reduction in hippocampal neuronal activation after most of the mTOR 3′-UTR (Terenzio et al. 2018). However, enriched environment exposure. This novel deletion meth- successful generation of an isoform-specific, long 3′-UTR odology has thus uncovered in vivo neuronal impairments knockout mouse using CRISPR–Cas9 has not been report- attributed to the loss of a single long 3′-UTR mRNA isoform ed to date. in mice. Calmodulin (CaM) is the primary calcium sensor in the cell (Means and Dedman 1980; Yamniuk and Vogel RESULTS 2004; Sorensen et al. 2013). CaM is expressed ubiquitously but is particularly abundant in the nervous system Calm1-L expression is enriched in neurons (Kakiuchi et al. 1982; Ikeshima et al. 1993). There are three Calmodulin genes in mammals—Calm1, Calm2, and To determine the relative expression of short and long Calm3. These share an identical amino acid coding se- Calm1 3′-UTR isoforms among mouse tissues, we per- quence, but possess unique 5′ and 3′-UTRs (SenGupta formed northern blot analysis—an approach uniquely suit- et al. 1987; Fischer et al. 1988), suggesting differences in ed to report on the expression of alternative 3′-UTRs (Miura their regulation might be conferred at the posttranscrip- et al. 2014). Northern blots performed using a universal tional level. The existence of alternative 3′-UTR mRNA iso- probe (uni) that detects both short and long Calm1 forms generated by APA for the Calmodulin 1 (Calm1) 3′-UTR isoforms revealed the presence of both Calm1-L gene have been known for several decades (SenGupta and Calm1-S across an array of adult tissues. The cerebral et al. 1987; Nojima 1989; Ni et al. 1992; Ikeshima et al. cortex showed greater enrichment of Calm1-L versus 1993). These include mRNAs with a short 0.9 kb 3′-UTR Calm1-S compared to non-brain tissues (Fig. 1A,B), as (Calm1-S) and one with a long 3.4 kb 3′-UTR (Calm1-L). had been previously observed (Ikeshima et al. 1993; The functional significance of these alternative 3′-UTR iso- Miura et al. 2013). The ratio of long isoform to the sum forms is entirely unexplored. of long and short isoforms (total Calm1) was found to be Previous work has shown that altering CaM levels dis- 0.46 in cortex, whereas in the rest of the tissues examined rupts neuronal development in Drosophila and rodents (gastrocnemius skeletal muscle, heart, testes, kidney, and (Vanberkum and Goodman 1995; Fritz and VanBerkum liver) this value ranged between 0.12–0.28. Recent studies 2000; Kim et al. 2001; Kobayashi et al. 2015; Wang et al. have shown that some 3′-UTRs can be cleaved to generate 2015). CaM plays a role in guiding axon projections stable 3′-UTR transcripts that are separated from the pro- to create connections with other cells (Vanberkum and tein coding portion of the mRNA (Kocabas et al. 2015; Goodman 1995; Fritz and VanBerkum 2000; Kim et al. Malka et al. 2017; Sudmant et al. 2018). To test whether 2001; Kobayashi et al. 2015). Additionally, CaM interplays this might be the case for Calm1, and to confirm the con- with a variety of synaptic proteins to play a fundamental nectivity of the long 3′-UTR region to the rest of the Calm1 role in regulating signaling pathways critical in synaptic mRNA, the blot was stripped and reprobed with a probe plasticity (Xia and Storm 2005). Functions in the nervous specific for Calm1-L (ext). One band consistent with the system specifically for Calm1 have been described. size of Calm1-L was observed (Fig. 1C). Thus, the long Notably, targeted knockdown of Calm1, but not Calm2 3′-UTR region is connected to the full-length mRNA, with or Calm3, was found to cause major migration defects in no evidence of a cleaved 3′-UTR observed in the tissues developing hindbrain neurons (Kobayashi et al. 2015). examined. Calm1 mRNA has been detected in cultured rat embryonic We next determined the spatial expression pattern of dorsal root ganglion (DRG) axons, without distinguishing Calm1 3′-UTR isoforms in mouse tissues. In situ hybridiza- Calm1-S or Calm1-L isoforms, where it undergoes local tion (ISH) using Digoxigenin (DIG)-labeled probes was per- translation to promote axon outgrowth in vitro (Wang formed in embryonic day 13.5 (E13.5) mice and 8-wk-old et al. 2015). Calm1 mRNAs, especially Calm1-L, were brains. Using a Calm1 universal probe (uni; Fig. 1A), strong found to be expressed in dendrites of rat hippocampal signal was observed in the brain (including forebrain [FB], neurons (Tushev et al. 2018). However, whether Calm1-L midbrain [MB], and hindbrain [HB]), spinal cord (SC), and

www.rnajournal.org 1415 Downloaded from rnajournal.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press

Bae et al.

A B C F

D E G

H I J

FIGURE 1. Calm1-L expression is enriched in neurons. (A) Diagram showing northern blot and DIG in situ probes to detect either both isoforms (uni) or specifically the long 3′-UTR isoform (ext). (B) Northern blot of an array of tissues collected from adult mice probed with uni probes showing Calm1-S (bottom arrow) and Calm1-L (top arrow). The ratio of Calm1-L normalized to total Calm1 (sum of Calm1-S band intensity and Calm1-L)is shown. (C) Northern blot performed with ext probe. The same blot was stripped and reprobed for the housekeeping gene Psmd4 as a loading control. (D) DIG in situ hybridization of E13.5 embryo and 8w brain using the uni probe showed universal expression of Calm1 showing strong signal in the nervous system including forebrain (FB), midbrain (MB), hindbrain (HB), spinal cord (SC), dorsal root ganglion (DRG), cortex, and hippocampus (Hpc). (E) DIG in situ performed with the ext probe showed neural tissue-specific expression pattern of Calm1-L with particularly strong signals in the DRG and Hpc. (F) RNA-seq tracks from purified neurons, microglia, astrocytes, newly formed oligodendrocytes, and endothelial cells are visualized. Calm1-L gene model is shown as annotated and Calm1-S gene model was illustrated based on the sequencing read cover- ages. Read coverage in the long 3′-UTR is particularly high in neurons. (G) QAPA was used to estimate the fraction of Calm1-S and Calm1-L found in each RNA-seq data set. The poly(A) usage indicates that Calm1-L is indeed enriched in neurons. Two replicates for each cell type. (H) Diagram showing RNAscope smFISH probe locations. (I,J) smFISH performed for total Calm1 (uni) or Calm1-L (ext) transcripts in primary DRG culture colabeled with β3 Tubulin (Tubb3) neuronal marker. Most cells showed robust uni signals while ext signals were restricted in Tubb3-positive cells. Significance was determined using a t-test, (∗∗∗∗) P < =0.0001. n = 58 Tubb3-pos cells, n = 23 Tubb3-neg cells. (K,L) smFISH performed in primary hippocampal culture showing the same trend. Significance was determined using a t-test, (∗∗∗∗) P < =0.0001. n = 17 Tubb3-pos cells, n = 9 Tubb3- neg cells. dorsal root ganglion (DRG) in parasagittal sections of em- DRG in parasagittal sections of the embryo and stronger bryos. Uni signal was also strong in whole coronal sections signal in the adult Hpc compared to other brain regions of the adult brain including amygdala, striatum, hypothal- (Fig. 1E). These experiments demonstrate that Calm1-L is amus, cortex, and hippocampus (Hpc) (Fig. 1D). Expres- enriched in neural tissues. sion was also observed in the nasal epithelium (N), lung The global bias for enhanced expression of longer 3′- (L), and intestinal epithelium (I). In contrast, ISH performed UTR isoforms in the nervous system has generally been at- with probe against the long 3′-UTR (ext; Fig. 1A) showed tributed to expression in neurons (Guvenek and Tian staining largely restricted to the brain, spinal cord, and 2018). To test if this is the case for Calm1, we analyzed

1416 RNA (2020) Vol. 26, No. 10 Downloaded from rnajournal.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press

Loss of Calm1-L impairs neuronal function the relative expression of the Calm1 3′-UTR isoforms in dif- puncta in these experiments due to colocalization of uni ferent brain cell types from a previously published RNA- (cyan) and ext (magenta) signals (Fig. 2A). We found that seq data set (Zhang et al. 2014). We performed reanalysis Calm1-L showed little to no signal in DRG axons (Fig. of these data using QAPA, which allows for an estimation 2B′,B′′). To confirm this expression pattern we reanalyzed of relative poly(A) site usage (PAU) from RNA-seq reads public RNA-seq data obtained from isolated axons of (Ha et al. 2018). We analyzed the usage of Calm1-S and DRG/spinal cord (Minis et al. 2014). QAPA analysis re- Calm1-L among purified neurons, microglia, astrocytes, vealed the relative expression of the long 3′-UTR to be newly formed oligodendrocytes, and endothelial cells. 23% in whole DRG. In isolated axons, there was less rela- The coverage tracks of these data showed the highest en- tive expression of the long 3′-UTR (11%) (Supplemental richment of reads pertaining to the long 3′-UTR in neurons Fig. 1a). This result is consistent with our FISH data show- compared to the other cell types (Fig. 1F). Quantification ing less enrichment of Calm1-L in axons (Fig. 2C). using QAPA revealed neurons had the greatest usage of A recent study that used global analysis of 3′ end se- Calm1-L (32%) (Fig. 1G). quencing found Calm1-L to be enriched over Calm1-S ex- To determine the cellular expression of Calm1 3′-UTR pression in the rat hippocampal neuropil (Tushev et al. isoforms in our tissues of interest, we performed 2018), which is a region that comprises axons, dendrites, branched-oligo based single molecule Fluorescence In synapses, interneurons, and glia (Cajigas et al. 2012). In or- Situ Hybridization (RNAscope smFISH) on primary cultures der to assess subcellular localization patterns of Calm1 3′- from DRG and hippocampus (Fig. 1H–L). Costaining with UTR isoforms in mouse hippocampal neurons, we cultured anti-β3Tubulin (Tubb3) was used to distinguish neurons postnatal day 0 (P0) mouse hippocampal neurons and per- from other cell types. In cultured DRG, most cells showed formed smFISH. β3Tubulin marker was colabeled to iden- robust uni signal, but ext signal was only prominent in tify neurons. We could not simultaneously monitor short Tubb3-positive cells (Fig. 1I). Counting of uni and ext and long isoforms along with multiple cellular markers puncta indicated that Calm1-L is mostly enriched in (for example, both β3Tubulin and Tau to distinguish axons Tubb3-positive neurons in DRG (Fig. 1J). Analysis of prima- and dendrites) by fluorescence microscopy, thus we refer ry hippocampal cultures similarly revealed that the Calm1 to all neuronal extensions as “processes” for this analysis. ext signal was significantly enriched in Tubb3-positive neu- The short 3′-UTR isoform was found in both soma and pro- rons (Fig. 1K,L). Together with the RNA-seq analysis of iso- cesses (cyan signals in Fig. 2D′′). In contrast to the results lated brain cells, these data show that Calm1-L expression observed in DRG neurons, Calm1-L was found to be local- is enriched in neurons. ized in hippocampal neuron processes (Fig. 2D′,D′′ white signals depicted with arrowheads). Quantification of the number of localized puncta per neuron and maximum dis- Subcellular localization of Calm1-L tance of travel for each isoform revealed no significant dif- Calm1 mRNA, without distinguishing between Calm1-S or ference between the two Calm1 isoforms (Fig. 2E; -L, has been shown to be localized to axons and dendrites Supplemental Fig. 1b). However, the distance of travel in multiple studies (Zivraj et al. 2010; Gumy et al. 2011; from the center of the soma out to processes was found Kobayashi et al. 2015; Wang et al. 2015; Preitner et al. to be significantly different between Calm1-S and 2016; Zappulo et al. 2017; Tushev et al. 2018). Given ear- Calm1-L, largely due to enriched Calm1-S distribution in lier work showing that elements present in long 3′-UTRs of the soma (Fig. 2F). Collectively, these results show that importin β1 and Impa1 are involved in mRNA localization Calm1-L is not expressed outside of the soma in DRG, to axons (Andreassi et al. 2010; Perry et al. 2012), we hy- but is expressed in both soma and processes in hippocam- pothesized that Calm1-L might be specifically localized pal neurons. to axons. To test this hypothesis, we cultured dissociated embryonic DRG neurons in compartmentalized chambers or on glass coverslips and performed smFISH. DRG neu- Generation of Calm1 long 3′′′′′-UTR deletion mice using rons are unipolar cells with a single axon stem bifurcating CRISPR–Cas9 into a peripheral and a central branch, and use of embryonic DRG allows a high efficiency of isolating neurons To investigate the functional role of neuron-enriched (Melli and Höke 2009). Previous studies have shown total Calm1-L in vivo, we used CRISPR–Cas9 gene editing to Calm1 expression in sensory neuron axons by multiple ap- generate a series of mouse lines that lacked the expression proaches, including FISH, microarray, and RNA-seq analy- of Calm1-L. We aimed to generate a mutant line which did sis of axonal transcriptomes (Zivraj et al. 2010; Gumy et al. not harbor any foreign DNA sequences to promote usage 2011; Wang et al. 2015; Preitner et al. 2016). In agreement of the proximal poly(A) site over the distal poly(A) site. We with these studies, we found that the probes targeting designed six single guide RNAs (gRNAs) targeting the both Calm1 isoforms (uni) revealed signal in soma and ax- Calm1 locus that we anticipated would result in three dif- ons (Fig. 2A,B). Calm1-L expression is indicated by white ferent deletions (Fig. 3A).

www.rnajournal.org 1417 Downloaded from rnajournal.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press

Bae et al.

A B C

B′

B″

E F D

D′

D″

FIGURE 2. Subcellular localization of Calm1-L.(A) Diagram showing RNAscope smFISH probe locations. When uni and ext images are merged, Calm1-L isoforms are shown as colocalized (white) puncta. (B) smFISH showed robust axonal localization of Calm1-S (uni), but Calm1-L (ext/magenta in B′ or white punta in B′′) was observed in DRG axons. (C) When the number of axonal uni or ext puncta was counted, uni signals were robustly found in the axons of DRG neurons but little to no ext signals were found in the same regions. Significance was determined using a t-test, (∗∗∗∗) P < =0.0001. n = 30 neurons. (D) In primary hippocampal neurons, both Calm1-S and Calm1-L were observed in soma and neuronal processes. (E) The number of puncta corresponding to Calm1-S and Calm1-L transcript in the hippocampal processes was not found to be significantly different. Significance was determined using a t-test, ns: P > 0.05. n = 15 neurons. (F) Analysis of the distance of travel for all the Calm1-S and Calm1-L signals showed the overall distribution is different between two mRNA isoforms. Analysis done in n = 15 neurons (Calm1-L 684 puncta, Calm1-S 907 puncta). Significance was determined using a Wilcoxon test, (∗∗∗) P < =0.001.

Deletion mice strains were generated by injection of all had the desired outcome. We found that Deletions 1 six gRNAs along with mRNA encoding Cas9 endonucle- and 3 exhibited complete loss of Calm-L transcripts ase. From 11 founder mice we isolated three lines each (Fig. 3B). As anticipated, northern analysis of deletion 2 harboring different deletions that we confirmed using line showed a band migrating slightly higher than the Sanger sequencing. Deletion 1 removed sequence en- Calm1-S band, indicating the biogenesis of an ectopic compassing the long 3′-UTR, including the distal poly(A) transcript slightly longer than Calm1-S due to the second site. Deletion 2 removed the majority of the sequence poly(A) signal (PAS) (Fig. 3B). Transcripts from the Calm2 comprising the long 3′-UTR, save for the distal poly(A) and Calm3 genes were also monitored by northern blot site. This strategy brought the proximal and distal poly(A) and were found to be unaltered (Fig. 3B; for full blots, sites adjacent to one another, which we surmised might see Supplemental Fig. 2). ensure cleavage at this region and prevent selection of In deciding on which line to carry out phenotypic analy- cryptic poly(A) sites further downstream. Deletion 3 was sis, we surmised that deletion 2 was the most confounding designed to remove the distal poly(A) site, in anticipation because an ectopic transcript was generated. The deletion that although the long 3′-UTR would be transcribed, it 1 and 3 strategies both had the desired effect on Calm1-L would not be cleaved and polyadenylated, thus prevent- loss without altering total Calm1 levels. We proceeded ing biogenesis of mature Calm1-L mRNA. with phenotypic analysis on the deletion 3 allele because The effectiveness in preventing Calm1-L biogenesis in it had the smallest amount of genomic sequence deleted these three lines was determined by northern blot analy- (164 bp) and thus reasoned that it was the least likely to sis using the uni probe (Fig. 1A). We chose to analyze cor- have alterations in genomic elements such as unknown en- tex samples instead of smaller tissues because a great hancer elements that could be present in the long 3′-UTR- deal of starting RNA is required to perform these encoding region. This mouse deletion line 3 was renamed Δ Δ mRNA northern blots. Remarkably, each deletion strategy Calm1 L/ L.

1418 RNA (2020) Vol. 26, No. 10 Downloaded from rnajournal.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press

Loss of Calm1-L impairs neuronal function

B C D

FIGURE 3. Generation of Calm1 long 3′-UTR deletion mice using CRISPR–Cas9. (A) Diagram of the strategies used to eliminate the production of mature long Calm1 3′-UTR transcripts. Six gRNAs (g1–g6) were injected simultaneously to generate a variety of deletions (Deletion 1–3). (B) Calm1 northern blot of adult cortex from three deletion strains and control littermate demonstrating successful deletion of the long Calm1 3′-UTR isoform. Note the Deletion 2 line generates a new isoform with a truncated long 3′-UTR due to the preservation of the distal PAS. Transcripts from the Δ Δ Calm2 and Calm3 genes were found to be unaltered. Psmd4 used as a loading control. (C) Complete loss of Calm1-L in the Calm1 L/ L (Deletion Δ Δ 3) hippocampal neurons has been confirmed by lacked smFISH signals representing Calm1-L.(D) DIG in situ performed in the Calm1 L/ L (Deletion 3) embryo also revealed no expression of Calm1-L.

We further verified the complete loss of Calm1-L in the which by early embryonic stage E10.5 have already begun Δ Δ Calm1 L/ L hippocampus using Stellaris smFISH and in to form distinct ganglia adjacent to the spinal neural tube the DRG by ISH. Cultured hippocampal neurons from caudal to the hindbrain (Fig. 4A; Le Douarin and Smith Δ Δ the Calm1 L/ L mice completely lacked signal represent- 1988; Marmigere and Ernfors 2007). Cell bodies of the ing Calm1-L (Fig. 3C). ISH for Calm1-L performed in DRG send out dorsolateral axon projections concurrent E13.5 embryo sections also revealed no expression (Fig. with neurogenesis during this stage in development 3D). We confirmed by Sanger sequencing that the (Marmigere and Ernfors 2007). Unlike DRG that form adult Δ Δ Calm1 L/ L line did not harbor unintended mutations at structures, the first cervical (C1) DRG is a temporary embry- predicted gRNA off target sites (Supplemental Fig. 3). onic population that loses its dorsolateral axonal projections and progressively undergoes programmed cell ΔL/ΔL death from E10.5 to ∼E12.5 (Fanarraga et al. 1997; van Calm1 mice exhibit DRG development den Akker et al. 1999). defects Δ Δ The axons and cell bodies of the C1 DRG in Calm1 L/ L Given the pronounced expression of Calm1-L in DRG (Fig. embryos at E10.5 were found to be severely disorganized Δ Δ 1E) we examined Calm1 L/ L embryos for DRG develop- relative to Calm1+/+ embryos (Fig. 4B,C). Large groups of mental defects. The DRG is a series of ganglia in the pe- cell bodies of the C1 DRG in mutant embryos translocat- ripheral nervous system derived from neural crest cells, ed rostral into the hindbrain adjacent to the accessory

www.rnajournal.org 1419 Downloaded from rnajournal.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press

Bae et al.

A BB′ B″ C C′ C″

D EF G

Δ Δ FIGURE 4. Calm1 L/ L mice exhibit DRG axon development defects. Developing C1 DRG exhibit axonal and cell body migration disorganization Δ Δ in Calm1 L/ L E10.5 embryos. (A) Schematic of Calm1+/+ E10.5 embryo highlighting the morphology of the C1 and C2 DRG axons and cell bodies. Δ Δ (B,C) Whole mount Calm1+/+ and Calm1 L/ L DRG morphology visualized by anti-Tubb3 labeling. (B) The cell bodies of the C1 DRG (arrowhead) in Calm1+/+ embryos are bundled together to form a distinct ganglion. The neurites of the C2 DRG can be seen projecting ventrally in an orga- Δ Δ nized bundle. (C) The C1 DRG in Calm1 L/ L animals is disorganized and consists of clusters of cell bodies (arrowheads) that extend bundles of axons (arrows). The C1 DRG cells of deletion animals migrate rostral relative to Calm1+/+ and are adjacent to the n.xi tract (B′,C′). Enlarged view of single optical section of same Z-stack from B and C focusing on disorganized mutant cell body morphology of ganglion that aberrantly migrated more rostral relative to control. (D–F) In order to start measurements in the same relative location between embryos, n.xii was used as an anatomical landmark to set a beginning of measurements, indicated by the gray dashed lines in B′′,C′′.(D) Quantification of area of ectopic cell bodies Δ Δ observed for developing DRG. (E) Distance of aberrantly clustered cell bodies in Calm1+/+ and Calm1 L/ L.(F) Quantification of the number of Δ Δ axon bundles projecting off C1 ganglia. Significance determined by a t-test; n = 4 C1 DRG Calm1+/+; n = 5 C1 DRG Calm1 L/ L. n.xi = accessory Δ Δ nerve, n.xii = hypoglossal nerve. (G) Capillary western analysis of Calm1+/+ and Calm1 L/ L embryonic DRG showing no change in overall protein levels. Significance was determined using a t-test, P = 0.802; n =3. nerve (n.xi) tracks (Fig. 4C–C′′). At the same location in were significantly higher numbers emanating from Δ Δ Calm1+/+ embryos, there were fewer translocating C1 Calm1 L/ L DRG compared to Calm1+/+ (Fig. 4F). DRG bodies as rostral and disorganized (Fig. 4B–B′′). Together these data show that loss of Calm1-L impairs Δ Δ The C1 DRG cell bodies in Calm1 L/ L grouped together C1 DRG axon development and restricts rostral cell in smaller ganglia (Fig. 4C–C′′). Axons branching off these migration. cell bodies projected aberrantly and were tightly fascicu- We reasoned that the impairment of DRG development Δ Δ Δ Δ lated (Fig. 4C′, see arrows). Some Calm1 L/ L axons pro- in Calm1 L/ L was a result of altered translation of Calm1 jected longitudinally, which was in contrast to the resulting from loss of the long 3′-UTR isoform. We thus per- dorsolateral projections seen in non-C1 DRG (see formed Western analysis for CaM in wild type and mutant Materials and Methods). Significantly more C1 DRG cell dissected E13.5 DRG. Note that three genes encode iden- bodies rostrally migrated in the mutants compared to tical CaM protein, thus this analysis cannot uniquely report controls (Fig. 4D,E). Lastly, we quantified the number of on CaM generated from Calm1. Western analysis was fascicles projecting off the C1 ganglia and found there performed using capillary immunoassay due to the low

1420 RNA (2020) Vol. 26, No. 10 Downloaded from rnajournal.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press

Loss of Calm1-L impairs neuronal function amounts of DRG material recovered from dissection. mRNAs in hippocampal neurons (Fig. 5C,D). Thus, our re- Western analysis failed to identify a significant change in sults suggested that the overall Calm1 mRNA levels, sub- Δ Δ CaM levels between Calm1+/+ and Calm1 L/ L embryonic cellular localization of Calm1 mRNAs, or CaM protein Δ Δ DRGs (Fig. 4G). levels are not significantly altered in Calm1 L/ L hippocampus compared to Calm1+/+.

Subcellular localization of Calm1 and CaM levels ΔL/ΔL ΔL/ΔL are unaltered in Calm1 hippocampus Calm1 mice exhibit reduced hippocampal IEG Δ Δ expression in response to enriched environment Next, we examined whether Calm1 L/ L mutants experi- exposure ence neurological defects in the central nervous system. We directed our attention to the hippocampus given the CaM is an integral part of CaMK, Calcineurin, and MAPK high expression of Calm1-L in this tissue (Fig. 1E). First, signaling pathways which are involved in synaptic trans- we characterized whether the loss of Calm1-L transcripts mission and plasticity (Xia and Storm 2005; Pang et al. altered the total amount of total Calm1 transcripts or 2010; Hagenston and Bading 2011; Sethna et al. 2016). CaM protein in the hippocampus. qRT-PCR and western Ca2+ signaling mediates synaptic plasticity via posttransla- blot analysis revealed no changes in total Calm1 RNA or tional alterations and initiating signaling cascades result- Δ Δ CaM protein levels between Calm1+/+ and Calm1 L/ L ing in de novo transcription (Greer and Greenberg 2008; (Fig. 5A,B). Additionally, we assessed changes in Calm1 Hagenston and Bading 2011). With this in mind, we hy- mRNA localization into the processes of hippocampal neu- pothesized the Calm1-L might impact CaM-dependent Δ Δ rons. smFISH analysis in both Calm1+/+ and Calm1 L/ L synaptic transmission and/or plasticity. As a proof of prin- hippocampal neurons showed that the deletion of ciple, we used an enriched environment (EE) paradigm Calm1-L did not significantly impair localization of Calm1 to induce experience-driven expression of immediate

A B

FIGURE 5. Characterization of Calm1 isoforms in the hippocampus. (A) qRT-PCR analysis revealed no changes in total Calm1 (uni) RNA levels Δ Δ Δ Δ between Calm1+/+ and Calm1 L/ L, while the Calm1-L expression is abolished in Calm1 L/ L hippocampus. Significance was determined using Δ Δ a t-test, ns: P > 0.05, (∗∗) P < =0.01, n = 4 mice. (B) Western blot analysis of Calm1+/+ and Calm1 L/ L adult hippocampus showing no change in overall protein levels. Significance was determined using a t-test, P = 0.287; n = 3 mice. (C,D) smFISH analysis in both Calm1+/+ and Δ Δ Calm1 L/ L hippocampal neurons showed that the deletion of Calm1-L did not significantly impair localizing potential of Calm1 mRNAs in hip- Δ Δ pocampal neurons. Significance was determined using a Wilcoxon test, ns: P > 0.05; n =19 Calm1+/+ neurons (739 puncta), n =26 Calm1 L/ L neurons (802 puncta).

www.rnajournal.org 1421 Downloaded from rnajournal.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press

Bae et al.

Δ Δ Δ Δ early genes (IEGs) in Calm1 L/ L hippocampi and com- reduced expression in the Calm1 L/ L hippocampus pared it to Calm1+/+. IEG expression, especially the canon- (Supplemental Fig. 4b). ical example of cFos induction, is a marker of neuronal The EE paradigm is influenced by multiple factors that activation and is considered to be required for learning mediate neural responses including social interaction, sen- and synaptic plasticity (Yap and Greenberg 2018). sory stimulation, learning, and motor activity (Alwis and Exposure to environmental novelty leads to a selective Rajan 2014). In order to exclude the possibility that the im- Δ Δ increase of cFos in the rat Cornu Ammonis 1 (CA1) region paired locomotor function of Calm1 L/ L mice might have (VanElzakker et al. 2008) and novel object and place induc- impacted EE induced IEG expression, we performed an Δ Δ es cFos expression in the rat CA1 (Ito and Schuman 2012). open field test. Both Calm1+/+ and Calm1 L/ L genotypes Immediately after exposing the mice to EE for 1.5 h, ani- showed similar levels of total ambulatory distance, mean mals were sacrificed, and brains were collected. EE-in- speed, time spent in periphery, and time spent in explora- duced expressions of cFos were compared in Calm1+/+ tion area (Fig. 6D). Thus, locomotor function and explor- Δ Δ and Calm1 L/ L CA1 (Fig. 6A–C). In home-cage control atory behavior were not altered by loss of Calm1-L under groups, <2% of cells in CA1 were cFos-positive. This is con- these conditions. sistent with previous reports that showed cFos expression We also performed CaM western blot of hippocampal Δ Δ in the home cage condition is extremely low, with numer- lysates from Calm1+/+ and Calm1 L/ L mice exposed to ous sections containing no positive neurons (Drew et al. the enriched environment (Supplemental Fig. 4c) and 2011). As expected, EE induced cFos expression in CA1 found total CaM protein levels to be unchanged. This re- to 31.9 ± 12.5% in Calm1+/+ mice (Fig. 6B,C). The EE in- sult was similar to what was observed in Jones et al., where Δ Δ duction of cFos expression in Calm1 L/ L mice was signifi- total CaM level in the hippocampus was unchanged in re- cantly reduced compared to wild-type controls (19.8 ± sponse to novel-context exposure (Jones et al. 2018). 9.1%) (Fig. 6B,C; Supplemental Fig. 4a). Analysis of Again, our results show that total CaM level was not altered Npas4, another IEG that is also activated in response to by the deletion of Calm1-L or by experience-dependent neuronal activity (Sun and Lin 2016), consistently showed neuronal activation.

A B C

FIGURE 6. Calm1-L loss reduces hippocampal IEG expression in response to enriched environment exposure. (A) Graphic illustration of enriched environment used in this study and the CA1 region shown in B.(B) Representative images showing EE-induced expressions of cFos in Calm1+/+ Δ Δ and Calm1 L/ L CA1. (C) Percentages of cFos-positive cells in whole CA1 region was quantified in FIJI/ImageJ. Significance was determined using a t-test, (∗∗∗) P < =0.001. n = 6 mice (two hemispheres in two brain sections for each mouse). (D) Levels of total distance traveled, mean speed, time Δ Δ spent in periphery and exploration areas of Calm1+/+ and Calm1 L/ L mice were compared using open field test. Significance was determined using a t-test in n = 5 mice; ns: P > 0.05.

1422 RNA (2020) Vol. 26, No. 10 Downloaded from rnajournal.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press

Loss of Calm1-L impairs neuronal function

Calm1-L is less stable than Calm1-S isoform while not altering the expression of its corresponding short 3′-UTR isoform. To our knowledge, this is the first Unable to pinpoint the exact molecular defect responsible successful implementation of such an approach specifi- for the phenotypes in DRG and hippocampus resulting ′ cally for a long 3 -UTR isoform in vivo. We found that elim- from Calm1-L loss, we turned our attention to the mRNA ′ ination of the Calm1 long 3 -UTR isoform impaired both stability of Calm1 3′-UTR isoforms. A commonly used tech- development of the DRG in embryos and activation of nique to estimate the half-life of mRNAs is tracing tran- adult hippocampal neurons, thus establishing a functional scripts by qRT-PCR during a timeframe after blocking role for Calm1-L in the peripheral and central nervous sys- transcription. Due to Calm1-S and Calm1-L isoforms shar- ′ tems. Our method for long 3 -UTR deletion using CRISPR– ing a common short 3′-UTR region, we cannot quantify Cas9 adds an important new tool for the characterization Calm1-S exclusively using qRT-PCR approaches in wild- ′ ΔL/ΔL of APA-generated long 3 -UTR transcript isoforms. type tissue. To overcome this, we used Calm1 mice ′ Despite the prevalence of alternative length 3 -UTR iso- to measure the stability of Calm1-S. Extension qRT-PCR forms in metazoan genomes, few physiological roles for primers (ext) were used to detect Calm1-L from Calm1+/+ these transcripts using loss-of-function genetic approach- samples, and universal qRT-PCR primers (uni) were used ΔL/ΔL es have been identified. In a previous study, an in vivo neu- to uniquely detect Calm1-S from Calm1 samples ′ rological function for the Bdnf long 3 -UTR isoform was (Supplemental Fig. 5). Hprt was quantified as a normalizing identified by generating a transgenic mouse that had tan- gene (La Fata et al. 2014). Primary cortical neurons were dem SV40 poly(A) sites inserted downstream from the treated with actinomycin D and samples collected immedi- proximal poly(A) signal to prevent biogenesis of the long ately and at 3, 6, and 8 h posttreatment. An exponential re- ′ 3 -UTR (An et al. 2008). Our strategy for generating loss gression equation was fitted to the relative abundance of ′ of long 3 -UTR isoforms is less confounding because artifi- each isoform across the timepoints as relative amount = − ∗ cial regulatory sequences are not inserted into the ge- e kdecay time and the half-life was estimated for each tran- nome. Including these strong regulatory sequences script. Calm1-S was found to have a longer half-life (t = ′ 1/2 generates chimeric short 3 -UTR transcripts, which can af- 5.9 ± 1.4 h) compared to Calm1-L (t = 2.6 ± 0.6 h) 1/2 fect cleavage and polyadenylation dynamics in unexpect- (Supplemental Fig. 5a). When Calm1 transcript levels ed ways. were normalized to the stable Hprt transcripts, Calm1-S We used multiple gRNAs for CRISPR–Cas9 gene editing was consistently found to be more stable than Calm1-L because it was not clear which, if any, of our deletion strat- (Supplemental Fig. 5b). To discard the possibility that ′ Δ Δ egies would effectively prevent Calm1 long 3 -UTR bio- Calm1 L/ L line might have overall RNA decay/stability genesis. One concern was that removal of the distal poly pathways affected due to the genomic manipulation, we (A) site might result in activation of downstream cryptic also measured the total Calm1 levels using uni primers in poly(A) sites. Remarkably, a single injection of this gRNA the Calm1+/+ samples. This again showed that the total cocktail led to three different deletions which all prevented Calm1 in Calm1+/+ samples has higher stability compared Calm1-L expression. We used mice generated by deletion to the Calm1-L (Supplemental Fig. 5c). Thus, we conclude strategy #3 for our analysis because this completely elimi- that Calm1-S isoform has a higher half-life compared to nated Calm1-L expression and only removed 164 bp of se- Calm1-L isoform. quence encompassing the distal poly(A) site. However, the Since enhanced RNA secondary structure affects mRNA other two deletion strategies might be worth considering half-life in mammalian systems (Sun et al. 2019), we exam- ′ for removing long 3 -UTR isoforms for other genes. For ex- ined predicted secondary structures in the long versus ′ ample, one advantage of deletion #1 strategy [which elim- short 3 -UTRs (Bellaousov et al. 2013). We calculated the ′ inates most of the long 3 -UTR and distal poly(A) site] is that length-normalized minimum thermodynamic free energy ′ it leaves no possibility for generating the long 3 -UTR iso- (−ΔG/nt) of predicted RNA structures and estimated the ′ form. However, implementing CRISPR for eliminating structural complexity of the 3 -UTRs in Calm1-S and ′ 3 -UTRs of extreme lengths (>10 kb) is likely to be of very Calm1-L (Supplemental Fig. 6). This analysis suggested low efficiency. Deletion strategy #2 removes the sequence more complex RNA structures are found in the Calm1-L in between the proximal and distal poly(A) sites. The major 3′-UTR (−ΔG/nt = −0.297) compared to the Calm1-S 3′- confounding factor for this strategy is that an ectopic, trun- UTR (−ΔG/nt = −0.248). Thus, differential RNA structure cated mRNA slightly longer than the short isoform is pro- could account for mRNA stability differences between duced due to the distal poly(A) site remaining intact. Calm1-S and Calm1-L. Still, this approach could be useful because bringing two adjacent poly(A) sites in proximity is likely to prevent usage of downstream cryptic poly(A) sites. DISCUSSION Genome-wide analyses consistently have found Calm1 Here, we successfully implemented CRISPR–Cas9 gene to be subcellularly localized in neurons (Gumy et al. editing in mice to eliminate expression of a long 3′-UTR 2011; Tushev et al. 2018). In accordance with these data,

www.rnajournal.org 1423 Downloaded from rnajournal.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press

Bae et al.

our smFISH experiments show Calm1 mRNA subcellular tivated synapse. Indeed, the activity-induced translation of localization to processes in both DRG and hippocampus. Calm1 has been supported by synaptoneurosome poly- A well characterized example of a gene with different sub- some fractionation experiments which showed a very small cellular localizations of 3′-UTR isoforms is Bdnf. Bdnf-L is but significant enrichment of Calm1 mRNAs in the heavy- dramatically enriched in dendrites compared to the soma polysome fraction after NMDAR-activation via NMDA and restricted Bdnf-S, and removal of the long 3′-UTR isoform glutamate (Jones et al. 2018). Further investigation into the was found to prevent mRNA localization into dendrites (An role of Calm1-L in activity-induced synaptic translation of et al. 2008). We had hypothesized that the long 3′-UTR of Calm1 would be enabled by generating a mouse harbor- Calm1 might serve a similar role—however, smFISH exper- ing an epitope tag fused to the Calm1 coding sequence Δ Δ iments revealed that both the short and long 3′-UTR iso- in the Calm1+/+ and Calm1 L/ L backgrounds. Combined forms were expressed in the processes of hippocampal with spatiotemporal analysis of translation using a method neurons, and only the short 3′-UTR isoform was found in such as Puro-PLA (Tom Dieck et al. 2015), such a transgenic DRG axons (Fig. 2). Thus, cis-elements uniquely found in line could be used to determine how Calm1-L 3′-UTR loss the long 3′-UTR of Calm1 do not drive the axon/dendrite impacts activity-induced CaM translation specifically from localization of Calm1 mRNAs. Along these lines, recent ge- the Calm1 locus in vivo. nome-wide analysis of transcript isoform localization in Finally, it is worth considering that the long 3′-UTR of somata versus neuropil similarly did not uncover a bias Calm1-L might not be particularly important for cis-regula- for longer 3′-UTR APA isoforms to be particularly localized tion of Calm1 translation. Instead of impacting translation to neuropil (Tushev et al. 2018). of their host genes, some long 3′-UTR isoforms act as scaf- Neural development requires execution of precise folds for assembling proteins that later form complexes gene regulatory programs in response to extracellular with the protein being translated (Brigidi et al. 2019; Lee cues, which results in cytoskeletal rearrangements allow- and Mayr 2019). Long 3′-UTRs have noncoding functions ing for the formation of neural circuits (Tessier-Lavigne that are completely removed from the functions of the and Goodman 1996). In recent years, gene regulation encoded protein. For example, long 3′-UTR isoforms of at the posttranscriptional level has been emerging as an Ube3a1 function as a miRNA sponge in the synaptoden- important aspect of axon growth and guidance (Holt dritic regions independent of its protein-coding ability, and Schuman 2013; Zhang et al. 2019a,b). In particular, and a nontranslatable long 3′-UTR isoform of Tp53inp2 is there are many examples of local translation occurring involved in TrkA receptor internalization in axons (Valluy in the developing growth cone in response to extracellu- et al. 2015; Crerar et al. 2019). lar cues (Campbell and Holt 2001; Lin and Holt 2007; The mechanism of how Calm1 long 3′-UTR isoform loss Zivraj et al. 2010; Shigeoka et al. 2013). Our data sug- impairs neuronal development and function remains un- gests that Calm1-L is important for C1 DRG cell body mi- clear. Nonetheless, our approach has uncovered neural gration and axon extensions during development. This phenotypes both in the PNS and CNS resulting from the was evident from in vivo experiments that displayed dra- loss of a long 3′-UTR isoform in mice. We have demonstrat- matically disorganized cell bodies and abnormal axonal ed an effective strategy for eliminating long 3′-UTR iso- Δ Δ extensions in the C1 population of DRG in Calm1 L/ L forms via CRISPR/Cas9 gene editing without affecting embryos (Fig. 4). Future studies are needed to determine short 3′-UTR expression. Thousands of alternative long how Calm1-L is involved in DRG cell body positioning 3′-UTR isoforms documented in mice are of unknown and axon outgrowth given that we did not observe a physiological relevance (Miura et al. 2013). The CRISPR ap- role for altered CaM protein nor altered axonal localiza- proach described here can be used to generate deletion tion of Calm1-L. strains on a gene-by-gene basis to determine which long EE-induced IEG expression was impaired in the hippo- 3′-UTR isoforms have in vivo functional relevance. Δ Δ campus of our Calm1 L/ L mice. IEGs play key roles in synaptic plasticity, social and cognitive functions (Coutellier et al. 2012; Chung 2015). It has been shown that blocking MATERIALS AND METHODS CaM activity inhibits long-term potentiation (Malenka et al. 1989), supporting a critical role of CaM in synaptic plastic- Animal use and tissue collection Δ Δ ity. It is thus possible that loss of Calm1-L in Calm1 L/ L mice impairs synaptic plasticity and/or cognitive functions. All mice were housed in an environmentally controlled facility under the supervision of trained laboratory support personnel. Despite the impairment in IEG activation, we found that Animal protocols were approved by the University of Nevada, CaM levels were unchanged in the hippocampus of Reno Institutional Animal Care and Use Committee (IACUC) and ΔL/ΔL Calm1 mice. It is possible that Calm1 is not regulated in accordance with the standards of the National Institutes of by activity-induced local translation. Alternatively, CaM Health Guide for the Care and Use of Laboratory Animals. translational changes resulting from Calm1-L loss might For embryo collection, crossed female mice were monitored be restricted to certain subcellular regions, such as the ac- daily for vaginal plugs, with positive identification being counted

1424 RNA (2020) Vol. 26, No. 10 Downloaded from rnajournal.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press

Loss of Calm1-L impairs neuronal function as E0.5 at noon that day. Pregnant mice were euthanized at noon RNA-seq analysis for 3′′′′′-UTR alternative using CO2 asphyxiation, and then cervical dislocation was per- polyadenylation identification formed. E10.5 and E13.5 embryos were extracted in PBS. Embryos were immediately used for fresh tissue collection or RNA-seq data sets for neuron, microglia, astrocyte, oligodendro- fixed in 4% PFA in PBS by immersion. Postnatal day 0 (P0)–P1 cyte, and endothelial cells were downloaded from GEO accession pups were euthanized by decapitation and tissues were collected series GSE52564, and data sets for DRG and isolated axons from in cold PBS or HBSS. For adult tissue RNA or protein extraction, GSE51572. Data sets were aligned and processed using HISAT2, dissected tissue was flash frozen in liquid nitrogen and stored at Samtools, and Bamtools. Bam or bedgraph files were loaded into ′ −80°C or immediately used. To obtain fixed tissues, 7- to 9-wk- IGV for track visualization. For 3 -UTRs alternative polyadenylation ′ old animals were euthanized by overdose of isoflurane inhalation analysis, 3 -UTRs reference was built according to QAPA algo- followed by transcardial perfusion with abundant PBS and 4% rithm with the exception of slight modification for Calm1 ′ ′ PFA in PBS. 3 -UTRs. The original QAPA reference contained 6 Calm1 3 - UTR isoforms, but unconfirmed shorter isoforms were omitted for accurate quantification of only Calm1-S and Calm1-L expressions. Isoform abundance was quantified using Sailfish and RT-qPCR and northern analysis QAPA quant algorithms. PAU values from the output file were used for comparison of 3′-UTR expression in each cell type. Flash frozen tissue was pulverized using a Cellcrusher tissue pul- verizer. RNA was then extracted using the RNeasy Plus Universal Mini Kit (Qiagen) or TRIzol (Qiagen) method and quan- Primary neuron culture tified using a NanoDrop spectrophotometer. For RT-qPCR, 1 or 2 µg of RNA was reverse transcribed using SuperScript III Reverse DRGs from E13.5 mouse embryos were dissected in Neurobasal Transcriptase (Invitrogen) or Maxima Reverse Transcriptase (Invi- medium. Cells were dissociated with 0.25% trypsin and triturated trogen). The cDNA reaction was diluted fivefold in ultrapure water with a gel-loading tip. Dissociated cells were plated on PDL and for use in RT-qPCR. RT-qPCR was performed using SYBR Select laminin coated compartmentalized chamber (Xona microfluidics, Master Mix for CFX (Applied Biosystems). The BioRad CFX96 XC450) or coverslips in culture medium (DMEM supplemented real time PCR machine was used to carry out real time PCR and with 1× Glutamax, 10% FBS, 25 ng/mL NGF, and 1× pen/strep). results were analyzed using the delta-delta CT method. For north- In particular for the compartmentalized chamber, the somal com- ern analysis, poly(A)+ RNA was extracted from total RNA using partment was coated only with Poly-D-lysine (PDL) and supple- NucleoTrap mRNA kit (Machery-Nagel). Northern blot analysis mented with medium containing 10 ng/mL NGF instead of was performed as previously described (Miura et al. 2013). Briefly, 25 ng/mL to promote the growing of axons through the micro- poly(A)+ RNA samples (2 µg) were denatured in glyoxal and run in groove and toward the axonal compartment. The cultures were BPTE gels prior to downward transfer followed by northern blot- maintained until DIV2 at 37°C with 5% CO2. ting using 32-P dCTP labeled DNA probes (sequences of primers Cortices or hippocampi from P0–P1 mouse pups were dissoci- used to generate probes are found in Supplemental Table 1). ated with 0.25% trypsin and plated in plating media (MEM sup- Blots were exposed overnight until desired intensity of signals plemented with 0.5% w/v glucose, 0.2 mg/mL NaHCO3, was detected using Typhoon FLA7000 phosphoimager (GE). 0.1 mg/mL transferrin, 10% FBS, 2 mM L-glutamine, and 0.025 mg/mL Insulin) onto PDL coated six-well plates. After 1 d in vitro (DIV), the media was replaced with growth media (hippocampal-

MEM supplemented with 0.5% w/v glucose, 0.2 mg/mL NaHCO3, Digoxigenin in situ hybridization 0.1 mg/mL transferrin, 5% FBS, 0.5 mM L-glutamine, and 2% B-27 supplement; cortical-Neurobasal media supplemented with 2% Riboprobes were generated via in vitro transcription using DIG B27 supplement and 1× GlutaMax). An amount of 4 µM AraC RNA labeling mix (Roche) for the same probe regions as in (Sigma-Aldrich) was added to the growth media at DIV1–2 to sup- northern blot analysis. Sucrose cryoprotected E13.5 embryos press the growth of nonneuronal cells and enrich postmitotic or adult brains were embedded in O.C.T compound and cryo- neurons. The cultures were maintained until DIV7 at 37°C with sectioned at 16–18 µm. Sections were treated in antigen retriev- 5% CO . al solution for 5 min in 95°C water bath and washed in water 2 twice. To aid permeabilization, slides were immersed in ice- cold 20% (v/v) acetic acid for 20 sec. Upon dehydration in Single molecule fluorescence in situ hybridization sequential washes in 70%–100% ethanol, slides were stored at −20°C until use. Approximately 30 ng riboprobes were used Single molecule FISH (smFISH) was performed in primary neurons in each hybridization at 65°C overnight. Stringent washes were using Stellaris (LGC Biosearch Technologies) or RNAscope multi- performed in SSC buffer (50% formamide in 2× SSC and 0.5× plex fluorescent assay according to manufacturer’s instruction. SSC). Anti-DIG-AP Fab antibody (Roche) incubation was per- Briefly, Stellaris smFISH was carried out according to manufactur- formed in 2% BSA in MABT solution for 1 h at room tempera- er’s protocol for adherent cells except that Wash buffer A was re- ture. Color development was carried out using NBT and BCIP placed with 2× SSC and 10% formamide in ultrapure water, (Roche) at room temperature until desired coloring was ob- Hybridization buffer with 2× SSC, 10% formamide, 10× dextran- served (between 16 to 20 h). Leica DM IL LED microscope or sulfate, and 250 nM probe. Hybridization was carried out for Keyence BZ-X710 all in one fluorescence microscope was used ∼18–24 h in the dark at 37°C. For RNAscope smFISH, cells were for imaging and stitched automatically. pretreated with permeabilization solution and protease solution,

www.rnajournal.org 1425 Downloaded from rnajournal.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press

Bae et al.

and incubated with 1× probe solution (551281-C3 or C1 for ex- Anti-CaM and anti-α-tubulin were used at 1:50 dilution. tension probe and 556541-C2 for universal probe diluted in Quantification of band intensity was automatically estimated us- probe diluent) for 2 h at 40°C. Subsequently, Amp1-FL to ing Compass for SW software. Amp4-FL were hybridized to amplify FISH signals. To combine FISH and immunofluorescence, immunostaining procedure was performed after the Amp4-FL hybridization step. Anti-Tubb3 pri- CRISPR–Cas9 gene editing mary and secondary antibodies were incubated for 1 h at 37°C The MIT CRISPR Design Tool (http://www.genome-engineering and RT, respectively. Samples were counterstained with .org/crispr) was used to design gRNAs targeted to various regions DAPI and mounted in anti-fade buffer (10 mM Tris pH 8.0, 2× of the Calm1 long 3′-UTR (Supplemental Table 1). Sequences SSC, and 0.4% glucose in water) or ProLong diamond antifade were cloned into the BbsI site of pX330-U6 chimeric BB-CBh- mountant (Invitrogen). During image acquisition, control neuron hSpCas9 plasmid (42230; Addgene). HiScribe T7 mRNA synthesis slides that had been incubated in probe diluent without probes kit (New England BioLabs) was used to in vitro transcribe the were used to set laser intensity to ensure no background signal guide RNAs, and RNAs were subsequently purified using RNA was counted in experimental conditions. Laser intensity and Clean & Concentrator-5 (Zymo Research, Cat. R1016) before as- detector range were kept constant among different conditions. sessment on an Agilent Bioanalyzer as described (Han et al. 2015). FISH-quant (Mueller et al. 2013) was used for puncta counting Super-ovulating 4–6-wk-old FVB/NJ female mice were mated of FISH signals and FIJI/ImageJ was used to measure the distance with C57BL/6J males. After fertilization, the eggs were collected of travel of each FISH signal from the center of the nucleus. from the oviducts. Mouse zygotes were microinjected into the cy- toplasm with Cas9 mRNA (100 ng/µL) and the gRNAs (100 ng/µL RNA stability assays and structure predictions each) as described previously (Han et al. 2015; Oliver et al. 2015; Halim et al. 2017). After injection, zygotes were cultured in KSOM

Cortical neurons were treated with 1 µg/mL Actinomycin D at + AA medium (Millipore) for 1 h at 5% CO2 before transfer into the DIV6. Total RNA was collected in TRIzol at 0, 3, 6, and 8 h post- 7- to 10-wk-old female CD1 foster mothers. treatment. An amount of 1 µg of RNA was reverse transcribed us- Genomic DNA was isolated from tail-snips of founder mice by ing Maxima Reverse Transcriptase (Invitrogen). RT-qPCR was overnight proteinase K digestion. Two sets of primers flanking performed as described above. To compare the relative stability the deletion regions were used for PCR genotyping to detect of Calm1-S and Calm1-L isoforms, expression levels of each tran- the deletion using Taq Polymerase (NEB). Sibling founder mice script for each time point were calculated relative to 0 h using were crossed together to generate the F1 generation. An F1 BioRad CFX Manager software and normalized to Hprt (a stable male harboring both the Deletion 2 and Deletion 3 alleles was transcript control). To estimate the half-life of each transcript, ex- crossed to C57BL/6 female mice to establish separate Deletion pression level for each time point was calculated relative to 0 h 2 and Deletion 3 lines. The F1 Deletion 1 animals were first without normalizing to Hprt. Exponential regression equations crossed to heterozygous littermates (Del1/WT x Del1/WT). An were fitted for each degradation plot. Half-life of each transcript F2 Del1 heterozygote was crossed to C57BL/6 female mice to es- was calculated using Goal-seek function in Excel to find the tablish the Deletion 1 line. All PCR products were resolved in 1% time point where the relative amount of transcript was reduced agarose gels. To confirm the deletions and check for any insertion to 50%. The half-life of each biological replicate was used to events, Sanger sequencing of gel purified PCR products was per- determine mean and standard deviation for each transcript. formed (Nevada Genomics Center, University of Nevada). For RNA structure prediction was performed in RNAstructure Web the experiments in this paper, Deletion 3 line was backcrossed server (Bellaousov et al. 2013) with default setting and −ΔG/nt with C57BL/6 at least three times to stabilize the background, was calculated using the energy value given from the analysis and heterozygote pairs were interbred to obtain Calm1+/+ and Δ Δ and the length of the 3′-UTR (Fischer et al. 2020). Calm1 L/ L unless otherwise stated.

Western analysis Immunofluorescence

For conventional western blot analysis, total protein was extracted For whole-mount analysis, small tears were made in embryos in in RIPA buffer supplemented with protease inhibitor tablet the forebrain and roof plate of the hindbrain to facilitate the pen- (Pierce). Protein samples were separated in 15% discontinuous etration of PFA into the neural tube. Embryos were fixed by over- SDS-PAGE gel and transferred onto 0.2 µm PVDF membrane night incubation in 4% PFA at 4°C. Embryos were blocked (Trans-Blot Turbo, BioRad). Membranes were blocked in 5% and permeabilized for antibody labeling by performing 3 × 5 skim milk followed by overnight incubation with primary antibod- min, 3 × 30 min, and then overnight washes of PBST (10% FBS ies at 4°C. Anti-CaM (Abcam 45689) and anti-α-tubulin (Sigma and 1% Triton X-100 in 1× PBS). Primary antibodies (Biolegend T9026) antibodies were used at 1:500–1000 and 1:2000, respec- cat#801201 anti-β3Tubulin; 1:1000) were then added and incu- tively. HRP conjugated secondary antibody incubation was per- bated for two nights rocking at 4°C. Embryos were again washed formed at room temperature for 1 h. HRP signal was detected for 3 × 5 min, 3 × 30 min, and then overnight in fresh PBST. using ProSignal Femto reagent (Prometheus) and imaged using Secondary antibodies were added at a 1:200 dilution for two a ChemiDoc Touch (Biorad). nights at 4°C rocking in the dark. Secondary antibodies were For low-input western blot analysis, capillary western system washed off again 3 × 5 min, 3 × 30 min, and then overnight in fresh (Wes, Protein simple) was used. A total of 500 ng of total protein PBST. Embryos were then immersed in 80% glycerol/PBS for a lysate was loaded per capillary to visualize proteins of interest. minimum of 2 h, were mounted, and then imaged whole mount.

1426 RNA (2020) Vol. 26, No. 10 Downloaded from rnajournal.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press

Loss of Calm1-L impairs neuronal function

A Leica SP8 TCS confocal microscope was used for imaging. For set through repetitive rounds of visual inspection that allowed the whole mount samples, images were taken at 1048 × 1048px, 200 most accurate detection of IEG positive cells and applied equally Hz scanning speed, 3-line averaging, 8 bit, and with the use of lin- across all the images. The number of IEG positive cells were then ear-z compensation. compared to the total DAPI counts in the same CA1 region to es- For free-floating brain section staining, PFA perfused brains timate the percentage of activated neurons. were postfixed in 1% PFA for two nights and sectioned using vibratome. An amount of 80 µm thick sections were immediately used or stored at −20°C in antifreeze solution (30% w/v sucrose, Open-field assay 1% w/v PVP-40, 30% v/v ethylene glycol in PBS) until use. To assess locomotor function and exploratory behavior of Sections were blocked in blocking solution for 2 h at RT, primary Δ Δ Calm1+/+ and Calm1 L/ L mice, open field assay was performed antibodies (cFos SYSY 226003; Npas4 Activity Signaling AS- using ANY-maze video tracking system. The day before the assay, AB18A) were incubated overnight at 4°C rocking in the dark. 10 age-matched naïve animals were placed in the behavior room in min × 3 washes were performed in PBS followed by fluorophore their home cage for at least 30 min and gently handled for a cou- labeled secondary antibody incubation at RT for 2 h. During 10 ple of minutes to minimize distress of animals during the test. The min × 3 washes, the nuclei were counterstained with DAPI. animals were returned to the housing room. On the test day, the Sections were mounted in ProLong diamond antifade mountant animals were brought in their home cages into the behavior room (Invitrogen). Sections were cured at RT for at least 3 h and stored 30 min prior to starting the test. Overhead lighting setup was used at 4°C until imaging. A Leica SP8 TCS confocal microscope was to prevent shadows. A single animal was placed in the periphery used for imaging at 1024 × 1024 px, 20× magnification, 3-line av- of the arena facing the wall, and its behavior was recorded for erage and images were stitched automatically. 10 min. The apparatus was cleaned with 70% ethanol and dried for 5 min between each testing session. Total distance traveled DRG migration in vivo analysis by each mouse, mean speed, time spent in periphery, and exploration area were obtained to compare the behavior between Δ Δ Whole mount E10.5 embryos were labeled with anti-β3Tubulin Calm1+/+ and Calm1 L/ L mice using ANY-maze software. (Biolegend Cat# 801201 1:1000) and were subsequently imaged by confocal microscopy. Images were analyzed using FIJI/ImageJ (NIH). The cell bodies to be quantified were defined by drawing a SUPPLEMENTAL MATERIAL line parallel to the last branch of the hypoglossal nerve extending across the length of the image. The hypoglossal nerve was chosen Supplemental material is available for this article. as an anatomical landmark because its morphology was unaffect- ed in mutant embryos and served as an independent reference ACKNOWLEDGMENTS point so measurements were consistent across embryos. The area and distance migrated of cell bodies and axons rostral to We thank members of the Miura laboratory, especially Vicente the hypoglossal branch were measured using a freehand selec- Gapuz III, Kayla Andrada, and Dr. Zhiping Zhang, for providing tion tool in FIJI/ImageJ. The multipoint cell counting tool was feedback and technical assistance with the project. We also thank used on FIJI/ImageJ to count the number of axon bundles Dr. Minkyung Kim for experimental advice and Dr. Brian Perrino branching off the C1 DRG. A two-sided Student’s t-test was per- and Dr. Brad Ferguson for sharing reagents. Funding was provid- formed to test for significance. ed to P.M. and W.Y. from National Institute of General Medical Sciences grant P30 GM110767; G.S.M. was supported by R01 EY025205 (National Eye Institute; P.M. was supported by R35 Enriched environment-induced neural activation GM138319 (National Institute of General Medical Sciences). assay Core facilities at the University of Nevada, Reno campus were supported by NIGMS COBRE P30 GM103650. Enriched environment (EE) was created in an 83.8 cm × 51.1 cm × H 34.3 cm box with sufficient bedding material at the bot- Received May 15, 2020; accepted June 6, 2020. tom. Different types of toys were placed in the box with the sides of the box covered with colorful papers as illustrated in Figure Δ Δ 6A. Age matched Calm1+/+ and Calm1 L/ L naïve animals (not REFERENCES used for other tests) were kept in their home cage undisturbed for at least 3 d prior to the experiment and transferred to the EE Alwis DS, Rajan R. 2014. Environmental enrichment and the sensory box on the test day. Mice were exposed to the EE for 1.5 h before brain: the role of enrichment in remediating brain injury. Front euthanasia and tissue fixation. Brain sections were analyzed for Syst Neurosci 8: 156. doi:10.3389/fnsys.2014.00156 An JJ, Gharami K, Liao G-Y, Woo NH, Lau AG, Vanevski F, Torre ER, IEG expression induction using free-floating immunofluores- ′ cence staining as described above. IEG positive cells were count- Jones KR, Feng Y, Lu B, et al. 2008. Distinct role of long 3 UTR BDNF mRNA in spine morphology and synaptic plasticity in ed in the entire CA1 regions using FIJI/ImageJ. Images were hippocampal neurons. Cell 134: 175–187. doi:10.1016/j.cell adjusted for levels and color balance automatically and pixels .2008.05.045 were smoothened. Then, image thresholding was performed us- Andreassi C, Zimmermann C, Mitter R, Fusco S, Devita S, Saiardi A, ing Huang’s method. After applying watershed function to sepa- Riccio A. 2010. An NGF-responsive element targets myo-inositol rate closely located cells, particle analysis was performed in CA1 monophosphatase-1 mRNA to sympathetic neuron axons. Nat area to count cells with pixel units above 40. The parameters were Neurosci 13: 291–301. doi:10.1038/nn.2486

www.rnajournal.org 1427 Downloaded from rnajournal.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press

Bae et al.

Bellaousov S, Reuter JS, Seetin MG, Mathews DH. 2013. Ha KCH, Blencowe BJ, Morris Q. 2018. QAPA: a new method for the RNAstructure: web servers for RNA secondary structure prediction systematic analysis of alternative polyadenylation from RNA-seq and analysis. Nucleic Acids Res 41: W471–W474. doi:10.1093/ data. Genome Biol 19: 45. doi:10.1186/s13059-018-1414-4 nar/gkt290 Hagenston AM, Bading H. 2011. Calcium signaling in synapse-to-nu- Blair JD, Hockemeyer D, Doudna JA, Bateup HS, Floor SN. 2017. cleus communication. Cold Spring Harb Perspect Biol 3: a004564. Widespread translational remodeling during human neuronal dif- doi:10.1101/cshperspect.a004564 ferentiation. Cell Rep 21: 2005–2016. doi:10.1016/j.celrep.2017 Halim D, Wilson MP, Oliver D, Brosens E, Verheij JB, Han Y, Nanda V, .10.095 Lyu Q, Doukas M, Stoop H, et al. 2017. Loss of LMOD1 impairs Brigidi GS, Hayes MGB, Delos Santos NP, Hartzell AL, Texari L, Lin PA, smooth muscle cytocontractility and causes megacystis microco- Bartlett A, Ecker JR, Benner C, Heinz S, et al. 2019. Genomic lon intestinal hypoperistalsis syndrome in humans and mice. decoding of neuronal depolarization by stimulus-specific Proc Natl Acad Sci 114: E2739–E2747. doi:10.1073/pnas NPAS4 heterodimers. Cell 179: 373–391.e27. doi:10.1016/j.cell .1620507114 .2019.09.004 Han Y, Slivano OJ, Christie CK, Cheng AW, Miano JM. 2015. CRISPR- Cajigas IJ, Tushev G, Will TJ, Tom Dieck S, Fuerst N, Schuman EM. Cas9 genome editing of a single regulatory element nearly abol- 2012. The local transcriptome in the synaptic neuropil revealed ishes target gene expression in mice–brief report. Arter Thromb by deep sequencing and high-resolution imaging. Neuron 74: Vasc Biol 35: 312–315. doi:10.1161/ATVBAHA.114.305017 – 453 466. doi:10.1016/j.neuron.2012.02.036 Holt CE, Schuman EM. 2013. The central dogma decentralized: new Campbell DS, Holt CE. 2001. Chemotropic responses of retinal perspectives on RNA function and local translation in neurons. growth cones mediated by rapid local protein synthesis and deg- Neuron 80: 648–657. doi:10.1016/j.neuron.2013.10.036 – radation. Neuron 32: 1013 1026. doi:10.1016/S0896-6273(01) Hoque M, Ji Z, Zheng D, Luo W, Li W, You B, Park JY, Yehia G, Tian B. 00551-7 2013. Analysis of alternative cleavage and polyadenylation by 3′ Chung L. 2015. A brief introduction to the transduction of neural ac- region extraction and deep sequencing. Nat Methods 10: 133– – tivity into Fos signal. Dev Reprod 19: 61 67. doi:10.12717/DR 139. doi:10.1038/nmeth.2288 .2015.19.2.061 Ikeshima H, Yuasa S, Matsuo K, Kawamura K, Hata J, Takano T. 1993. Coutellier L, Beraki S, Ardestani PM, Saw NL, Shamloo M. 2012. Expression of three nonallelic genes coding calmodulin exhibits Npas4: a neuronal transcription factor with a key role in social similar localization on the central nervous system of adult rats. J and cognitive functions relevant to developmental disorders. Neurosci Res 36: 111–119. doi:10.1002/jnr.490360112 PLoS One 7: e46604. doi:10.1371/journal.pone.0046604 Ito HT, Schuman EM. 2012. Functional division of hippocampal area Crerar H, Scott-solomon E, Bodkin-clarke C, Gaspari M, Kuruvilla R, CA1 via modulatory gating of entorhinal cortical inputs. Crerar H, Scott-solomon E, Bodkin-clarke C, Andreassi C, Hippocampus 22: 372–387. doi:10.1002/hipo.20909 Hazbon M, et al. 2019. Regulation of NGF signaling by an axonal Jones KJ, Templet S, Zemoura K, Kuzniewska B, Pena FX, Hwang H, untranslated mRNA report regulation of NGF signaling by an axo- Lei DJ, Haensgen H, Nguyen S, Saenz C, et al. 2018. Rapid, expe- nal untranslated mRNA. Neuron 102: 553–563.e8. doi:10.1016/j rience-dependent translation of neurogranin enables memory en- .neuron.2019.02.011 coding. Proc Natl Acad Sci 115: E5805–E5814. doi:10.1073/pnas Drew LJ, Stark KL, Fénelon K, Karayiorgou M, Macdermott AB, .1716750115 Gogos JA. 2011. Evidence for altered hippocampal function in a Kakiuchi S, Yasuda S, Yamazaki R, Teshima Y, Kanda K, Kakiuchi R, mouse model of the human 22q11.2 microdeletion. Mol Cell Sobue K. 1982. Quantitative determinations of calmodulin in the Neurosci 47: 293–305. doi:10.1016/j.mcn.2011.05.008 supernatant and particulate fractions of mammalian tissues. J Fanarraga ML, Charite J, Hage WJ, De Graaff W, Deschamps J. 1997. Biochem 92: 1041–1048. doi:10.1093/oxfordjournals.jbchem Hoxb-8 gain-of-function transgenic mice exhibit alterations in the peripheral nervous system. J Neurosci Methods 71: 11–18. doi:10 .a134019 .1016/S0165-0270(96)00122-7 Kim YS, Furman S, Sink H, VanBerkum MF. 2001. Calmodulin and pro- – Fischer R, Koller M, Flura M, Mathews S, Strehler-Page MA, Krebs J, filin coregulate axon outgrowth in Drosophila. J Neurobiol 47: 26 Penniston JT, Carafoli E, Strehler EE. 1988. Multiple divergent 38. doi:10.1002/neu.1013 mRNAs code for a single human calmodulin. J Biol Chem 263: Kobayashi H, Saragai S, Naito A, Ichio K, Kawauchi D, Murakami F. 17055–17062. 2015. Calm1 signaling pathway is essential for the migration of – Fischer JW, Busa VF, Shao Y, Leung AKL. 2020. Structure-mediated mouse precerebellar neurons. Development 142: 375 384. RNA decay by UPF1 and G3BP1. Mol Cell 78: 70–84.e6. doi:10 doi:10.1242/dev.112680 Kocabas A, Duarte T, Kumar S, Hynes MA. 2015. Widespread differen- .1016/j.molcel.2020.01.021 ′ Floor SN, Doudna JA. 2016. Tunable protein synthesis by transcript tial expression of coding region and 3 UTR sequences in neurons – isoforms in human cells. Elife 5: e10921. doi:10.7554/eLife.10921 and other tissues. Neuron 88: 1149 1156. doi:10.1016/j.neuron Fritz JL, VanBerkum MF. 2000. Calmodulin and son of sevenless de- .2015.10.048 pendent signaling pathways regulate midline crossing of axons La Fata G, Gärtner A, Domínguez-Iturza N, Dresselaers T, Dawitz J, in the Drosophila CNS. Development 127: 1991–2000. Poorthuis RB, Averna M, Himmelreich U, Meredith RM, Achsel T, Greer PL, Greenberg ME. 2008. From synapse to nucleus: calcium-de- et al. 2014. FMRP regulates multipolar to bipolar transition affect- pendent gene transcription in the control of synapse development ing neuronal migration and cortical circuitry. Nat Neurosci 17: and function. Neuron 59: 846–860. doi:10.1016/j.neuron.2008.09 1693–1700. doi:10.1038/nn.3870 .002 Le Douarin NM, Smith J. 1988. Development of the peripheral ner- Gumy LF, Yeo GS, Tung YC, Zivraj KH, Willis D, Coppola G, Lam BY, vous system from the neural crest. Annu Rev Cell Biol 4: 375– Twiss JL, Holt CE, Fawcett JW. 2011. Transcriptome analysis of 404. doi:10.1146/annurev.cb.04.110188.002111 embryonic and adult sensory axons reveals changes in mRNA rep- Lee SH, Mayr C. 2019. Gain of additional BIRC3 protein functions ′ ertoire localization. RNA 17: 85–98. doi:10.1261/rna.2386111 through 3 -UTR-mediated protein complex formation. Mol Cell Guvenek A, Tian B. 2018. Analysis of alternative cleavage and polya- 74: 701–712.e9. doi:10.1016/j.molcel.2019.03.006 denylation in mature and differentiating neurons using RNA-seq Lianoglou S, Garg V, Yang JL, Leslie CS, Mayr C. 2013. Ubiquitously data. Quant Biol 6: 253–266. doi:10.1007/s40484-018-0148-3 transcribed genes use alternative polyadenylation to achieve

1428 RNA (2020) Vol. 26, No. 10 Downloaded from rnajournal.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press

Loss of Calm1-L impairs neuronal function

tissue-specific expression. Genes Dev 27: 2380–2396. doi:10 Preitner N, Quan J, Li X, Nielsen FC, Flanagan JG. 2016. IMP2 axonal .1101/gad.229328.113 localization, RNA interactome, and function in the development of Liao GY, An JJ, Gharami K, Waterhouse EG, Vanevski F, Jones KR, axon trajectories. Development 143: 2753–2759. doi:10.1242/ Xu B. 2012. Dendritically targeted Bdnf mRNA is essential for en- dev.128348 ergy balance and response to leptin. Nat Med 18: 564–571. doi:10 Sandberg R, Neilson JR, Sarma A, Sharp PA, Burge CB. 2008. .1038/nm.2687 Proliferating cells express mRNAs with shortened 3′ untranslated Lin AC, Holt CE. 2007. Local translation and directional steering in ax- regions and fewer microRNA target sites. Science 320: 1643– ons. EMBO J 26: 3729–3736. doi:10.1038/sj.emboj.7601808 1647. doi:10.1126/science.1155390 Malenka RC, Kauer JA, Perkel DJ, Mauk MD, Kelly PT, Nicoll RA, SenGupta B, Friedberg F, Detera-Wadleigh SD. 1987. Molecular anal- Waxham MN. 1989. An essential role for postsynaptic calmodulin ysis of human and rat calmodulin complementary DNA clones. and protein kinase activity in long-term potentiation. Nature 340: Evidence for additional active genes in these species. J Biol 554–557. doi:10.1038/340554a0 Chem 262: 16663–16670. Malka Y, Steiman-Shimony A, Rosenthal E, Argaman L, Cohen- Sethna F, Zhang M, Kaphzan H, Klann E, Autio D, Cox CL, Wang H. Daniel L, Arbib E, Margalit H, Kaplan T, Berger M. 2017. Post-tran- 2016. Calmodulin activity regulates group I metabotropic gluta- scriptional 3´-UTR cleavage of mRNA transcripts generates thou- mate receptor- mediated signal transduction and synaptic depres- sands of stable uncapped autonomous RNA fragments. Nat sion. J Neurosci Res 94: 401–408. doi:10.1002/jnr.23719 Commun 8: 2029. doi:10.1038/s41467-017-02099-7 Shigeoka T, Lu B, Holt CE. 2013. RNA-based mechanisms underlying Marmigere F, Ernfors P. 2007. Specification and connectivity of neuro- axon guidance. J Cell Biol 202: 991–999. doi:10.1083/jcb – nal subtypes in the sensory lineage. Nat Rev Neurosci 8: 114 127. .201305139 doi:10.1038/nrn2057 ′ Smibert P, Miura P, Westholm JO, Shenker S, May G, Duff MO, Mayr C, Bartel DP. 2009. Widespread shortening of 3 UTRs by alterna- Zhang D, Eads BD, Carlson J, Brown JB, et al. 2012. Global pat- tive cleavage and polyadenylation activates oncogenes in cancer terns of tissue-specific alternative polyadenylation in Drosophila. – cells. Cell 138: 673 684. doi:10.1016/j.cell.2009.06.016 Cell Rep 1: 277–289. doi:10.1016/j.celrep.2012.01.001 – Means AR, Dedman JR. 1980. Calmodulin an intracellular calcium re- Sorensen AB, Sondergaard MT, Overgaard MT. 2013. Calmodulin in a – ceptor. Nature 285: 73 77. doi:10.1038/285073a0 heartbeat. FEBS J 280: 5511–5532. doi:10.1111/febs.12337 Melli G, Höke A. 2009. Dorsal root ganglia sensory neuronal cultures: Sudmant PH, Lee H, Dominguez D, Heiman M, Burge CB. 2018. a tool for drug discovery for peripheral neuropathies. Expert Opin Widespread accumulation of ribosome-associated isolated Drug Discov 4: 1035–1045. doi:10.1517/17460440903266829 ′ 3 UTRs in neuronal cell populations of the aging brain. Cell Rep Miller S, Yasuda M, Coats JK, Jones Y, Martone ME, Mayford M. 2002. 25: 2447–2456.e4. doi:10.1016/j.celrep.2018.10.094 Disruption of dendritic translation of CaMKIIα impairs stabilization Sun X, Lin Y. 2016. Npas4: linking neuronal activity to memory. Trends of synaptic plasticity and memory consolidation. Neuron 36: 507– Neurosci 39: 264–275. doi:10.1016/j.tins.2016.02.003 519. doi:10.1016/S0896-6273(02)00978-9 Sun L, Fazal FM, Li P, Broughton JP, Lee B, Tang L, Huang W, Kool ET, Minis A, Dahary D, Manor O, Leshkowitz D, Pilpel Y, Yaron A. 2014. Chang HY, Zhang QC. 2019. RNA structure maps across mamma- Subcellular transcriptomics-dissection of the mRNA composition lian cellular compartments. Nat Struct Mol Biol 26: 322–330. in the axonal compartment of sensory neurons. Dev Neurobiol doi:10.1038/s41594-019-0200-7 74: 365–381. doi:10.1002/dneu.22140 Terenzio M, Koley S, Samra N, Rishal I, Zhao Q, Sahoo PK, Urisman A, Miura P, Shenker S, Andreu-Agullo C, Westholm JO, Lai EC. 2013. ′ Marvaldi L, Oses-Prieto JA, Forester C, et al. 2018. Locally translat- Widespread and extensive lengthening of 3 UTRs in the mamma- ed mTOR controls axonal local translation in nerve injury. Science lian brain. Genome Res 23: 812–825. doi:10.1101/gr.146886.112 359: 1416–1421. doi:10.1126/science.aan1053 Miura P, Sanfilippo P, Shenker S, Lai EC. 2014. Alternative polyadenylation in the nervous system: to what lengths will 3′ UTR extensions Tessier-Lavigne M, Goodman CS. 1996. The molecular biology of – take us? Bioessays 36: 766–777. doi:10.1002/bies.201300174 axon guidance. Science 274: 1123 1133. doi:10.1126/science Mueller F, Senecal A, Tantale K, Marie-Nelly H, Ly N, Collin O, .274.5290.1123 Basyuk E, Bertrand E, Darzacq X, Zimmer C. 2013. FISH-quant: au- Tom Dieck S, Kochen L, Hanus C, Heumüller M, Bartnik I, Nassim- tomatic counting of transcripts in 3D FISH images. Nat Methods Assir B, Merk K, Mosler T, Garg S, Bunse S, et al. 2015. Direct visu- 10: 277–278. doi:10.1038/nmeth.2406 alization of newly synthesized target proteins in situ. Nat Methods – Ni B, Rush S, Gurd JW, Brown IR. 1992. Molecular cloning of calmod- 12: 411 414. doi:10.1038/nmeth.3319 Tushev G, Glock C, Heumuller M, Biever A, Jovanovic M, ulin mRNA species which are preferentially expressed in neurons ′ in the rat brain. Brain Res Mol Brain Res 13: 7–17. doi:10.1016/ Schuman EM. 2018. Alternative 3 UTRs modify the localization, 0169-328X(92)90039-E regulatory potential, stability, and plasticity of mRNAs in neuronal – Nojima H. 1989. Structural organization of multiple rat calmodulin compartments. Neuron 98: 495 511. doi:10.1016/j.neuron.2018 genes. J Mol Biol 208: 269–282. doi:10.1016/0022-2836(89) .03.030 90388-4 Ulitsky I, Shkumatava A, Jan CH, Subtelny AO, Koppstein D, Bell GW, Oliver D, Yuan S, McSwiggin H, Yan W. 2015. Pervasive genotypic mo- Sive H, Bartel DP. 2012. Extensive alternative polyadenylation dur- saicism in founder mice derived from genome editing through ing zebrafish development. Genome Res 22: 2054–2066. doi:10 pronuclear injection. PLoS ONE 10: e0129457. doi:10.1371/jour .1101/gr.139733.112 nal.pone.0129457 Valluy J, Bicker S, Aksoy-Aksel A, Lackinger M, Sumer S, Fiore R, Pang ZP, Cao P, Xu W, Südhof TC. 2010. Calmodulin controls synaptic Wüst T, Seffer D, Metge F, Dieterich C, et al. 2015. A coding-inde- strength via presynaptic activation of calmodulin kinase II. J pendent function of an alternative Ube3a transcript during neuro- Neurosci 30: 4132–4142. doi:10.1523/JNEUROSCI.3129-09.2010 nal development. Nat Neurosci 18: 666–673. doi:10.1038/nn Perry RBT, Doron-Mandel E, Iavnilovitch E, Rishal I, Dagan SY, .3996 Tsoory M, Coppola G, McDonald MK, Gomes C, Vanberkum MFA, Goodman CS. 1995. Targeted disruption of Ca2+- Geschwind DH, et al. 2012. Subcellular knockout of importin β1 calmodulin signaling in Drosophila growth cones leads to stalls perturbs axonal retrograde signaling. Neuron 75: 294–305. in axon extension and errors in axon guidance. Neuron 14: 43– doi:10.1016/j.neuron.2012.05.033 56. doi:10.1016/0896-6273(95)90239-2

www.rnajournal.org 1429 Downloaded from rnajournal.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press

Bae et al.

van den Akker E, Reijnen M, Korving J, Brouwer A, Yap EL, Greenberg ME. 2018. Activity-regulated transcription: bridg- Meijlink F, Deschamps J. 1999. Targeted inactivation of Hoxb8 af- ing the gap between neural activity and behavior. Neuron 100: fects survival of a spinal ganglion and causes aberrant limb 330–348. doi:10.1016/j.neuron.2018.10.013 reflexes. Mech Dev 89: 103–114. doi:10.1016/S0925-4773(99) Zappulo A, van den Bruck D, Ciolli Mattioli C, Franke V, Imami K, 00212-9 McShane E, Moreno-Estelles M, Calviello L, Filipchyk A, VanElzakker M, Fevurly RD, Breindel T, Spencer RL. 2008. Peguero-Sanchez E, et al. 2017. RNA localization is a key determi- Environmental novelty is associated with a selective increase in nant of neurite-enriched proteome. Nat Commun 8: 583. doi:10 Fos expression in the output elements of the hippocampal forma- .1038/s41467-017-00690-6 tion and the perirhinal cortex. Learn Mem 15: 899–908. doi:10 Zhang Y, Chen K, Sloan SA, Bennett ML, Scholze AR, O’Keeffe S, .1101/lm.1196508 Phatnani HP, Guarnieri P, Caneda C, Ruderisch N, et al. 2014. Wang H, Yang H, Shivalila CS, Dawlaty MM, Cheng AW, Zhang F, An RNA-sequencing transcriptome and splicing database of glia, Jaenisch R. 2013. One-step generation of mice carrying mutations neurons, and vascular cells of the cerebral cortex. J Neurosci 34: in multiple genes by CRISPR/Cas-mediated genome engineering. 11929–11947. doi:10.1523/JNEUROSCI.1860-14.2014 Cell 153: 910–918. doi:10.1016/j.cell.2013.04.025 Zhang M, Ergin V, Lin L, Stork C, Chen L, Zheng S. 2019a. Wang B, Pan L, Wei M, Wang Q, Liu W-W, Wang N, Jiang X-Y, Axonogenesis is coordinated by neuron-specific alternative splic- Zhang X, Bao L. 2015. FMRP-mediated axonal delivery of miR- ing programming and splicing regulator PTBP2. Neuron 101: 690– 181d regulates axon elongation by locally targeting Map1b and 706.e10. doi:10.1016/j.neuron.2019.01.022 Calm1. Cell Rep 13: 2794–2807. doi:10.1016/j.celrep.2015.11 Zhang Z, So K, Peterson R, Bauer M, Ng H, Zhang Y, Kim JH, Kidd T, .057 Miura P. 2019b. Elav-mediated exon skipping and alternative poly- Xia Z, Storm DR. 2005. The role of calmodulin as a signal integrator for adenylation of the Dscam1 gene are required for axon outgrowth. synaptic plasticity. Nat Rev Neurosci 6: 267–276. doi:10.1038/ Cell Rep 27: 3808–3817.e7. doi:10.1016/j.celrep.2019.05.083 nrn1647 Zivraj KH, Tung YCL, Piper M, Gumy L, Fawcett JW, Yeo GSH, Holt CE. Yamniuk AP, Vogel HJ. 2004. Calmodulin’s flexibility allows for pro- 2010. Subcellular profiling reveals distinct and developmentally miscuity in its interactions with target proteins and peptides. Mol regulated repertoire of growth cone mRNAs. J Neurosci 30: Biotechnol 27: 33–57. doi:10.1385/MB:27:1:33 15464–15478. doi:10.1523/JNEUROSCI.1800-10.2010

1430 RNA (2020) Vol. 26, No. 10 Downloaded from rnajournal.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press

Elimination of Calm1 long 3′-UTR mRNA isoform by CRISPR−Cas9 gene editing impairs dorsal root ganglion development and hippocampal neuron activation in mice

Bongmin Bae, Hannah N. Gruner, Maebh Lynch, et al.

RNA 2020 26: 1414-1430 originally published online June 10, 2020 Access the most recent version at doi:10.1261/rna.076430.120

Supplemental http://rnajournal.cshlp.org/content/suppl/2020/06/10/rna.076430.120.DC1 Material

References This article cites 84 articles, 20 of which can be accessed free at: http://rnajournal.cshlp.org/content/26/10/1414.full.html#ref-list-1

Open Access Freely available online through the RNA Open Access option.

Creative This article, published in RNA, is available undera Creative Commons License Commons (Attribution 4.0 International), as described at License http://creativecommons.org/licenses/by/4.0/.

Email Alerting Receive free email alerts when new articles cite this article - sign up in the box at the Service top right corner of the article or click here.

To subscribe to RNA go to: http://rnajournal.cshlp.org/subscriptions